The hidden cost of coding errors

Quantifying the downstream impact

Coding errors tend to show up in conversations about claim denials as that is the most visible consequence. But denial rates capture only part of what inaccurate coding actually costs a health system. The full picture is larger, less visible, and considerably harder to reverse.

What do the numbers say

The scale of the problem is documented at the federal level. CMS reported that Medicare Fee-for-Service paid $28.83 billion in improper payments in fiscal year 2025, representing a 6.55% improper payment rate. The single most common driver across settings was insufficient documentation: missing records, documentation that did not support the level of service billed, and coding that could not be verified against the clinical record.

Medicare Part C, which covers Medicare Advantage plans, recorded an improper payment rate of 6.09%, amounting to $23.67 billion. CMS noted that most Part C improper payments were attributable to situations where the Medicare Advantage Organization's supporting documentation failed to substantiate the beneficiary diagnosis data submitted for payment. That is a documentation and coding problem, not a billing one.

The coding intensity issue in Medicare Advantage runs deeper still. MedPAC estimates that the federal government pays $84 billion more annually for Medicare Advantage enrollees than it would pay for similar patients in traditional Medicare, with $40 billion of that gap attributable to coding differences that increase risk scores and payments to MA plans. That figure reflects the systemic financial consequence of coding applied unevenly, imprecisely, or without adequate documentation to support it.

The two directions errors run

Coding errors are often discussed as overcoding, the risk of billing for more than the clinical record supports. That risk is real and carries regulatory consequences. But undercoding is equally costly and far less discussed.

When a chronic condition goes uncaptured, a comorbidity is missing from the record, or a hierarchical condition category is not coded because documentation is ambiguous, the health system does not get paid for the complexity of care it actually delivered. In value-based and risk-adjusted payment models, HCC capture failure reduces risk scores and capitation payments directly, depressing revenue not just on a single claim but across an entire patient population. That leakage is quiet, invisible in standard denial reports, and almost never recovered.

HFMA has noted that errors in HCC coding and missing comorbidities often go unnoticed initially but are quickly flagged by payers during retrospective review, creating audit exposure after the fact for revenue that was already recognized and spent.

The pattern is the same in both directions. Overcoding creates repayment liability. Undercoding creates permanent revenue loss. Both trace back to the same root: a translation process that is inconsistent, rules-dependent, and difficult to audit after the fact.

The operational cost that does not appear in revenue reports

Beyond the direct financial impact, coding errors create a secondary cost that is harder to measure but just as real. Every denied claim generates a workflow. Someone has to identify the denial, research the root cause, correct the underlying documentation or code, and resubmit. Across a health system handling thousands of encounters per week, that rework consumes substantial staff time and produces no new revenue. It is recovery work, not productive work.

As noted in article two of this series, HFMA has reported that hospitals lose an average of 4.8% of net revenue to denials, and that a significant share of denied claims are never resubmitted at all. The cost of that abandonment is pure write-off, compounded by the administrative overhead already spent on the initial submission and review.

There is also a compliance cost that sits outside the revenue cycle entirely. When documentation does not support the codes submitted, health systems face audit exposure from CMS, OIG, and commercial payers simultaneously. Preparing for and responding to those audits requires dedicated resources, and findings can result in repayment obligations that dwarf the original revenue at stake.

Why this is a coding architecture problem

The common response to coding errors is more training, more audits, and more review. Those interventions help at the margins. They do not address the underlying structural issue: that a process relying on individual judgment applied consistently across thousands of encounters, under time pressure, against a ruleset that changes every year, will produce errors at a predictable rate regardless of how skilled the coders are.

Accuracy at scale requires a different approach. It requires coding logic that is applied consistently across every encounter, with every rule applied the same way each time, and output that documents its own reasoning so it can be audited without additional reconstruction. That is what removes the gap between what was coded and what the clinical record supports.

The cost of coding errors is not abstract. It shows up in denied claims, in underpaid risk scores, in audit findings, and in the operational overhead of managing all of the above. Getting the translation right the first time is not an efficiency play. It is a financial one.

Next in this series: A developer's guide to integrating medical coding capabilities.